Meet the team: ECMWF
Digital twins of the Earth system are at the heart of Destination Earth (DestinE). They allow us to simulate our planet in more detail and test possible what-if scenarios, helping us better prepare for extreme events and develop more effective strategies for adapting to a changing climate.
Given ECMWF’s experience with numerical weather prediction and climate modelling, the European Commission tasked the institution with implementing DestinE’s Digital Twins. The organisation is also implementing the Digital Twin Engine, a modular software infrastructure that enables the digital twins to run on the EuroHPC supercomputers and handle the vast amounts of data. The digital twins are focused on climate change adaptation and weather-induced extremes and the organisation also focuses on a range of AI activities.
To achieve this broad range of activities, ECMWFEuropean Centre for Medium-Range Weather Forecasts More works with more than 100 partner organisations across Europe.Meet some of the people from ECMWFEuropean Centre for Medium-Range Weather Forecasts More who are working closely with these partner organisations to make it all happen:
Two digital twins, hundreds of developers
Sebastian Milinski is a senior scientist at ECMWFEuropean Centre for Medium-Range Weather Forecasts More contributing to the Climate Change Adaptation Digital Twin (Climate DT), which is implemented by CSC in collaboration with ECMWFEuropean Centre for Medium-Range Weather Forecasts More. With a background in climate modelling, he coordinates Climate DT related activities at ECMWFEuropean Centre for Medium-Range Weather Forecasts More, as well as ECMWF’s close collaboration with the CSC led partnership. His role includes technical coordination – ensuring smooth operation of the Climate DT – and scientific leadership, helping to define future priorities and directions for further evolution.
Together with his colleagues, he ensures the system is developed with users in mind by connecting modelling experts with national weather and climate services that understand the needs of those requiring climate information.
“We’re doing something that hasn’t been done before, which is to operationalise climate projections,” he says. “Our work builds on decades of expertise in Numerical Weather Prediction, where operational forecasts have been run daily for years, and now we’re extending that proven approach to climate projections.”
Benoît Vannière is a scientist at ECMWFEuropean Centre for Medium-Range Weather Forecasts More for the Weather-Induced Extremes Digital Twin (Extremes DT). The Extremes DT consists of two components: a global component developed at ECMWFEuropean Centre for Medium-Range Weather Forecasts More and a regional one implemented by a partnership led by Meteo-France that includes many national meteorological services across Europe.
Vannière, whose background is in climate modelling, coordinates the simulations produced within the global component at ECMWFEuropean Centre for Medium-Range Weather Forecasts More. He oversees its configuration, ensuring that simulation sets are robust and properly evaluated and contributes to running these simulations on Europe’s high-performance computing infrastructure. He also coordinates the scientific collaboration with the Meteo-France led partnership that implements the regional component.
“In a changing climate, extreme weather events will become more frequent. With the Extremes DT, we’re aiming to develop an additional tool to explore extreme events in past, present and future climates,” Vannière says.
Making the digital twins accessible and AI-ready
Mathilde Leuridan is a research software engineer at ECMWFEuropean Centre for Medium-Range Weather Forecasts More. She works on parts of the Digital Twin Engine, a central component of DestinE. The Engine brings together the software infrastructure necessary for extreme-scale simulations, data handling, data access and machine learning. “I mostly work on the digital twin data once it has been produced,” she says. “I work on the systems in-between, and make the petabytes of generated data available to users in a more intuitive and efficient way.”
Leuridan significantly contributes to developing PolytopeRESTful datacube access service for flexible, highly perform More, ECMWF’s user-facing data service for DestinE Digital Twin data. This functionality enables users to retrieve only the bytes of data they need from the vast volumes of weather and climate data produced by the Digital Twins. “It’s been really exciting for me to see how the innovative software that we’re developing can really help users make the most of the data and drive impact in fields like climate and weather prediction,” she adds.
Sara Hahner is a machine learning (ML) scientist at ECMWFEuropean Centre for Medium-Range Weather Forecasts More. With a background in mathematics and computer science, she contributes to building ML models for Earth system components such as waves, land, ocean, sea ice and hydrology, in the framework of DestinE. This complements efforts at ECMWFEuropean Centre for Medium-Range Weather Forecasts More to build AIFS, the Artificial IntelligenceArtificial Intelligence is the capacity of an algorithm to a More/Integrated Forecasting System. This is an AI-based weather forecasting system that uses machine learning to predict the atmosphere faster and more efficiently than traditional physics-based models.
By coupling the ML components developed in DestinE with AIFS, which mainly focuses on information from the atmosphere, the goal is to build a fully coupled AI Earth system model. To build these models, Hahner and her colleagues begin by working closely with domain scientists to understand the physics of the component they are working on. With the information she gets, she then develops and trains machine learning models to predict the evolution of that Earth system component.
To Hahner, one important milestone was to see that the ML wave model she worked on successfully captured extreme events. In December 2024, a North Pacific storm generated waves travelling across the ocean for several days. It was particularly meaningful to Hahner when their wave model captured the evolution of this event that caused devastating impacts.
“Being able to apply machine learning to something so tangible and relevant to people’s lives is deeply rewarding,” she says. “I’m driven by the idea that our work could help many people better understand and prepare for what’s coming.
If you would like to find out more about the digital twins, you can read ECMWF’s explainers and visit the DestinE Platform to access DT data.
To keep up with all DestinE news, follow us on LinkedIn. And to receive updates directly in your inbox, subscribe to our community newsletter.